Avoidant and resistant reunion behaviors as predicted by maternal interactive behavior and infant temperament

Avoidant and resistant reunion behaviors as predicted by maternal interactive behavior and infant temperament

INFANT BEHAVIOR AND DEVELOPMENT 12, 105-117 (19B9) Avoidant and Resistant Reunion Behaviors as Predicted by Maternal Interactive Behavior and I...

912KB Sizes 1 Downloads 22 Views

INFANT

BEHAVIOR

AND

DEVELOPMENT

12,

105-117

(19B9)

Avoidant and Resistant Reunion Behaviors as Predicted by Maternal Interactive Behavior and Infant Temperament GUNILLA BOHLIN, BERIT HAGEKULL, MARIT GERMER, KERSTIN ANDERSSON, AND LENE LINDBERG University of Uppsala, Sweden The study concerned predictive relationships from maternal behavior and infant temperament, studied at infant age 4 months, to infant resistant and avoidant reunion behaviors at 12 months. Subjects were 116 mother-infant pairs. Maternal interactive behaviors were observed in both face-to-face and diaper change situations, and infant ll-month behaviors were observed in a separation-reunion series, tailored after the laboratory Strange Situation, performed at home. Significant predictive relationships were found from measures related to the maternal sensitivity construct. Stepwise regression analyses showed differential predictions to hold for three variables a priori defined as components of sensitivityinsensitivity. Avoidant infant behavior was predicted by a combination of maternal responsiveness and physical contact, whereas infant resistant behavior was predicted by maternal intrusiveness. With forced entry of the Ainsworth global rating of sensitivity after the three components, the global rating was found to contribute significantly to the prediction of resistant, but not avoidant. reunion behavior. None of the dimensions of infant temperament correlated significantly with later infant avoidant or resistant behavior, but the interaction between maternal physical contact and infant intensity/activity contributed significantly to the prediction of infant avoidant behavior.

infant resistant

maternal behavior

sensitivity avoidant

temperament behavior

The notion of maternal sensitivity as the core aspect of infants’ early interactional experiences has been emphasized in attachment theory and research. The empirical basis was originally provided by Ainsworth, Blehar, Waters, and Wall (1978) in their pioneering work on the development of attachment. They showed that the maternal behaviors observed early in infancy, which later on distinguished securely attached from insecurely attached children, were related This research was supported by grants from the Swedish Council for Research in the Humanities and Social Sciences. Correspondence and requests for reprints should be sent to Gunilla Bohlin, Department of Clinical Psychology, University of Uppsala, Box 1225, S-751 42 Uppsala, Sweden. 105

106

BOHLIN,

HAGEKULL,

GERMER,

ANDERSON,

AND

LINDBERG

to the mother’s ability to perceive and consistently and adequately respond to her infant’s needs and communications. The concept of maternal sensitivity has come to hold a central position in attempts to understand early socioemotional development, but critical voices have also been raised. One line of critique stems from the field of temperament research, where it has been argued that the emphasis on infants’ interactional experiences overlooks the importance of early infant characteristics in an understanding of later individual differences (e.g., Goldsmith & Alansky, 1987). Generally, studies have shown little direct predictive value of infant temperament variables for the classification of secure and insecure attachments (Belsky & Rovine, 1987; Sroufe, 1985), but the possibility of interactive effects (i.e., the idea that, only in combination with certain maternal behaviors, specific infant characteristics might lead to a specific attachment outcome; cf. Sroufe, 1985) merits further study. Another point of criticism regarding sensitivity (e.g., Lamb, Thompson, Gardner, & Charnov, 1985) concerns the conceptual problem that the unidimensional construct of maternal sensitivity is considered predictive of three different outcomes: secure, avoidant, and resistant attachment. Thus, insensitive mothering is related to two different kinds of insecure child behavior, but the conditions leading to one or the other are not specified. Although one strength of the sensitivity concept is that it encompasses many aspects of caregiver behavior, it might be fruitful to distinguish salient components of it, both to further the understanding of its conceptual meaning and to aid in the search for specific antecedents of the two categories of anxious attachment. The core aspect of sensitivity, as defined by Ainsworth, Bell, and Stayton (1971), no doubt is responsiveness to infant signals and communications, which has been empirically defined in terms of responses to infant social and distress behaviors (e.g., Belsky, Rovine, & Taylor, 1984; Blehar, Lieberman, & Ainsworth, 1977). A second component which might be important is emotional physical contact. Such behavior has been considered to have a specific relationship to the avoidant category of anxious attachment (e.g., Main & Weston, 1982). A third aspect of sensitivity-insensitivity entails active maternal behavior which is not geared to infant signals (cf. Ainsworth et al., 1971; Belsky et al., 1984). In contrast to unresponsiveness, which is generally characterized by passivity or ignorance, this component is experienced by an observer as an overdoing of the interaction. Already Bowlby (1969) suggested that both “too little” and “too much” mothering may lead to anxious attachment, and Bohlin and Hagekull(l987) found a score of “too much” of certain maternal interactive behaviors to be negatively related to concomitant adequate infant behavior. This third aspect of maternal sensitivity, which might be denoted intrusiveness, encompasses behavioral features such as overstimulation, interference with infant activity, and bad timing and intensity-modulation of maternal behaviors. Thus, it also captures important aspects of the global rating of maternal cooperation-interference used by Ainsworth and collabora-

ANTECEDENTS

OF INFANT

REUNION

BEHAVIORS

107

tors, which was found to distinguish between securely and insecurely attached groups (Ainsworth et al., 1978). In the present study, the general aim was to study predictive relationships from maternal behavior and infant temperament during early infancy to later infant behaviors. More specifically, the empirical questions concerned whether responsiveness, physical contact, and intrusiveness could be described as constituent components of the more global construct maternal sensitivity-insensitivity, their predictive relationships to later infant behavior in a separation-reunion situation, and the predictive power of early temperament, both alone and in combination with maternal sensitivity, to the same outcome measures. In studying the predictive relationships, the outcome measures chosen were obtained from the attachment tradition. A standardized home observation was used, involving a series of separations and reunions, similar to those of the laboratory Strange Situation, developed to assess attachment security (Ainsworth et al., 1978). The variables studied in the home situation were resistant and avoidant behavior during reunions, that is, the behaviors that most clearly characterize the two categories of insecure attachment, when observed in the standard Strange Situation. METHOD Subjects The study was part of a larger ongoing longitudinal project in which families were enrolled when their infant was 6 weeks of age. Of the mothers contacted, I23 (61%) agreed to participate (for details about selection, sample characteristics, and outline of the total longitudinal study, the reader is referred to the first author). Maternal age ranged from 19 to 44 years (M= 29.8 years; SD =4.7 years). The level of education was comparatively high with 50% having some kind of university education. Fourteen percent had started or completed other post-high school training, 28% had completed high school, and 8% had only the compulsory 9 years of schooling. Two thirds of the families lived in suburbs or the outskirts of Uppsala, Sweden, 21% lived in the central parts, and 13% lived in rural areas. All but two of the mothers lived with the infant’s father. The target child was the firstborn child in 48 families; in the other 75 families, there were one to three siblings (M= 1.47). Of the participating infants, 63 were male and 60 were female. When the infants were 4 months old (M= 17.5 weeks, SD = 0.8 weeks), 119 families participated in the data collection, Two families had moved, one mother objected to the contents of the interviews, and one mother found that the questionnaires were too difficult to fill in. When the infants were 12 months of age (M= 54.7 weeks; SD = 2.2 weeks), 116 mother-infant pairs participated in the data collection. The reasons for nonparticipation were lack of time, problems getting a babysitter for siblings during the planned home visit, and illness in the family.

108

BOHLIN,

HAGEKULL,

GERMER,

ANDERSON,

AND

LINDBERG

Procedure TWO weeks before the 4-month visit mothers received a mailed infant temperament questionnaire. At infant age 4 months, two observers visited 74 and 45 homes each, collected the filled-out questionnaires, interviewed the mothers, and made observations of mother-infant interaction in three different situations: face-to-face interaction, diaper change, and feeding of the infant. The duration of the face-to-face situation was 5 min; the other two observations lasted until the task was completed. Mothers were told to behave as naturally as possible and to make their own choice for the order of the three observations. Data from the feeding situation were obtained with a different set of observational items than was used in the other situations and will be reported elsewhere. The 12-month home visit encompassed observations of infant sociability, infant behavior in a separation-reunion series, and mother-infant interaction during free play, performed in that order. Only data from the separationreunion series will be reported here. Three women psychologists, unfamiliar with attachment theory in general and the study hypothesis in particular, visited 29, 63, and 24 homes each. The families were seen by the same investigator during the first phases of data collection in an effort to create a feeling of consistency and familiarity in the 4-year longitudinal study, thereby minimizing attrition in the sample. For the present data set, therefore, the same person met with 75 families for the 4- and the 12-month visits. In view of the multitude of data gathered and the theoretical ignorance of the investigators, it was considered unlikely that any theoretically meaningful results would be produced by observer bias. The fact that interobserver reliability for the 1Zmonth observation was good supports the argument against the existence of halo effects from the 4-month observations in the present analyses. Temperament data were unknown to the observers. The separation-reunion series consisted of four 3-min phases: a separation, a reunion, a second separation and reunion. The sociability observation, lasting less than 5 min (Stevenson & Lamb, 1979), directly preceded the first separation, and the separation was thus initiated in a situation where the stranger was interacting with the child, which is similar to the preseparation episode in the laboratory Strange Situation. The separation and reunion phases were tailored as closely as possible to the procedure described by Ainsworth et al. (1978). Before the first separation, the mother was instructed to leave the room, close the door, and stay outside until the observer signaled by knocking (after 3 min). The mother was told that she could return before the knock if she wanted to. She was also told to call out the child’s name before opening the door and to pause on the doorstep to allow the child to see that she was coming back. During the first separation, the observer awaited the child’s immediate search behavior before starting an interaction. The observer then geared her behavior to that of the infant and tried to relieve distress by soothing or dis-

ANTECEDENTS

OF INFANT

REUNION

BEHAVIORS

109

tracting. Before the mother’s re-entrance, interaction was stopped and the observer moved away from the child. During reunion, mothers were free to take whatever course of action they wanted. Time for reunion was scheduled to 3 min but could be prolonged if the child was still distressed. Then the mother was again told to leave, to close the door, and to await the return signal after 3 min (or return on her own initiative). To mirror the second separation in the standard Strange Situation (when the infant is left alone), the stranger/ observer was now passive and did not try to attract the child’s interest or attention. However, if the child communicated to her, she answered adequately. The second reunion was performed like the first one. Methods and Measures Infant temperament was measured with the Baby Behavior Questionnaire (BBQ), which contains 54 statements about infant everyday behavior in a wide range of situations to be responded to on 5-step scales with verbally defined endpoints. Using a representative sample (n = 791) of Swedish infants, the instrument was developed within the behavioral style framework (Thomas, Chess, Birch, Hertzig, & Korn, 1963) from the Carey Infant Temperament Questionnaire (ITQ; Carey, 1970). From a series of factor analyses, including crossvalidation, six scales were derived, measuring the dimensions of Intensity/Activity, Regularity, Approach-Withdrawal, Sensory Sensitivity, Attentiveness, and Manageability (Bohlin, Hagekull, & Lindhagen, 1981; Hagekull & Bohlin, 1981). The issue of parental reports as valid measures of infant characteristics was attacked in a series of studies in which the empirical validity of the scales was established as correlating with direct observations of infant behavior (Hagekull, Bohlin, 8z Lindhagen, 1984). Estimated coefficients of concurrent validity were above average validity estimates presented for other parent-report instruments. Accounts of the unexplained variance recognized problems with both the behavioral criterion chosen and the parent questionnaire technique. It was, however, concluded that substantial portions of BBQ variability (between 28% and 72% for the different scales; M=44%) could be attributed to infant behavioral variation. Adequate test-retest reliability and homogeneity of the scales has been reported from previous samples (Hagekull & Bohlin, 1981); scale homogeneity as expressed by alpha coefficients in the present sample was a close replication of previous alpha values. The present coefficients ranged between .63 and .77 for the scales, with the exception of Manageability (ar = .50; for a discussion of the developmental course of the Manageability dimension, see Hagekull, in press). During the mother-infant face-to-face and diaper change interactions at infant age 4 months, behaviors were coded according to a translated and slightly modified version of the method developed by Vaughn, Taraldson, Crichton, and &eland (1980). Regarding maternal behavior, the scheme included an overall judgment of maternal sensitivity to infant signals and communications

110

BOHLIN,

HAGEKULL,

GERMER,

ANDERSSON.

AND

LINDBERG

(as defined by Ainsworth et al., 1971) and 12 specific items tapping the amount and quality of interactive behavior. The main change from the scheme used by Vaughn et al. concerned the inclusion of the item Intrusiveness, which was defined as overwhelming the-infant with stimulation, interfering with infant activity, or unduly hindering or interrupting the infant’s behavior and expressions. Thus, this item has much in common with the global characteristic cooperation-interference, as described by Ainsworth et al. (1971), but emphasizes overstimulation, that is, “too much” of otherwise positive interactive behaviors. Eight infant behavior items all dealt with social initiative and responsiveness, Each item was rated on a 5-step scale. In the present study, only the maternal variables a priori defined as potential predictors for subsequent child reunion behaviors were used. The variables were Sensitivity, Physical Contact, Responsiveness, and Intrusiveness in the main analyses averaged across the face-to-face and the diaper change situations. Sensitivity was considered a general aspect of maternal interactive behavior, and the other three variables were considered constituent components of it, although not necessarily exhausting its conceptual meaning. The measures of Sensitivity (scale endpoints: insensitive-very sensitive) and Intrusiveness (scale endpoints: never-often) were obtained as the scores given on the respective items. A measure of Physical Contact was obtained as the average score of the items measuring amount of emotional body contact (scale endpoints: none-almost constant) and quality of body contact (face-to-face scale endpoints: rigid/avoids-adaptive; diaper change scale endpoints: uncertain/ impatient-skilled/adaptive). For Responsiveness, the intention was to average the items concerning frequency of maternal answers to infant’s social initiatives, latency of response to infant distress, and effectiveness in comforting. Because infant distress occurred in less than half of the sample, the latter two items had to be dropped, and the Responsiveness measure consisted of the score on the first item, frequency of answers to infant’s social initiatives (scale endpoints: never-always). The remaining items concerned amount of verbal communications, amount of eye contact/seeking of eye contact, amount of expressions of positive emotions, amount of expressions of negative emotions, amount of attention to infant, and intensity of positive responses (variables not used here). Before starting the observations, observers were trained by rating video recordings of mother-infant pairs not belonging to the study sample. Interobserver reliability was obtained from parallel observations of the study sample and yielded r( 17) = .73 for Sensitivity, r( 17) = .75 for Physical Contact, r( 17) = .87 for Intrusiveness, and r(16) = .79 for Responsiveness. No significant differences in observer means were obtained. The observations during the two separation and reunion episodes at the 12month home visit were coded on an observation sheet, containing the various behaviors described in Ainsworth et al. ‘s(1978) scoring manual. During the

ANTECEDENTS

OF INFANT

REUNION

111

BEHAVIORS

observation, behaviors were marked on the coding sheet and these markings were then transformed to ratings on 7-point scales according to the manual. In the present analyses, data from the scales Avoidant Behavior and Resistant Behavior were used. The variable Resistant Behavior was the average of the scores obtained in the two reunions. For Avoidant Behavior, the rating in the second reunion was used because in the laboratory Strange Situation avoidant behavior during the second reunion has been found to be more discriminative in distinguishing between the secure and the insecure-avoidant categories than avoidance in the first reunion; some securely attached children typically show some avoidance during the first reunion, but this diminishes in the second reunion (Ainsworth et al., 1978). In the present study, no conclusion was changed when using avoidance scores averaged across reunions, but the results were generally weaker. Observers were trained to identify relevant behaviors by the first author (who received her training with professor L. Alan Sroufe) with the use of pilot study video recordings of Strange Situations performed in the laboratory with infants not in the study sample. Interobserver reliability, calculated on parallel observations spread over the whole data collection periods, amounted to r(17) = .94 for resistance averaged over the two reunions and r(17) = 1 .O for avoidance during the second reunion. RESULTS Components of Maternal Sensitivity Table 1 presents the intercorrelations between the behavioral items considered to be components of sensitivity and also their relations to the global rating. The correlation between responsiveness and physical contact is low but significant, while intrusiveness is unrelated to both of them. All three specific behaviors correlated significantly with the global rating. This is also true for the partial correlations, showing the relationship between a specific behavior and the global rating, with control of the other two specific items. To further verify TABLE 1 lntercorreiotions Among Meosures of Maternal (the Three Separote Components of Sensitivity Physical Responsiveness Physical contact Intrusiveness

the

contact0 .lB*

Interactive Behaviors and the Global Rating)

Intrusivenessa -.09 -36

Note, Values in parentheses are partial correlations between global rating with the other two components controlled for. ’ df=l13-116. b Partial correaltion: df=lll. * p<.o5. ** p<.Ol. *** p<.ool.

Global

Sensitivityaob

.30** .24”

(.34***)

.42’*’ each

of the

(.29**) (so-)

components and

112

BOHLIN,

HAGEKULL,

GERMER,

ANDERSSON,

AND

LINDBERG

that the maternal variables Responsiveness, Physical Contact, and Intrusiveness could all be seen as separate components of the more global construct sensitivity-insensitivity, a stepwise multiple regression analysis was performed with the global rating of Sensitivity as the dependent variable and the three presumed components as independent variables. All three components yielded significant contributions to a multiple R of .58 (R2= .32 when adjusted for sample size and number of variables). To examine whether the global rating of sensitivity tapped the same aspects of maternal behavior in the two observation situations, separate regression analyses on diaper change and face-to-face data were also performed. All three components entered with significant contributions in both situations, however, the first variable entered for the face-to-face situation was Intrusiveness, whereas the first variable for the diaper change situation was Responsiveness. The multiple Rs were .52 for the diaper change and .69 for the face-to-face situation. Predictive Relationships To evaluate the predictive power of the three components of sensitivity, multiple regression analyses with stepwise inclusion of independent variables were performed with resistant and avoidant child behavior in the reunion episodes as dependent variables. Because the global aspect of sensitivity could not be considered exhaustively described by the specific items, the global sensitivity rating was forced into the regression after the stepwise inclusion of Responsiveness, Physical Contact, and Intrusiveness, thereby evaluating the predictive power of unspecified aspects of Sensitivity. The results, which are presented in Table 2, show that the two categories of insecure reunion behavior were differentially predicted by the various sensitivity components. Whereas resistant behavior was significantly predicted by maternal intrusiveness, avoidant beTABLE Regression

Statistics From

Predictor

Criterion

Step

Resistant

1 2

Intrusiveness Sensitivity

Avoidant

1 2 3

Responsiveness Physical contact Sensitivity

(global

(global Note. corrected l

2

for Predictions of Observed Observed Maternal Voriables

Infant Reunion at Infant Age

Behaviors 4 Months

at 12 Months

F for R’ Change

Equotion

df

B

F for

R

R’odi

.21 .30

.04 .OB

5.03’ 5.83’

5.03’ 5.54”

1,110 2,109

.21 -.24

.30 .37

.00 .12

10.17** 5.81 l

10.17” B.22”*

.30

.12

0.63

5.67*’

1,106 2,105 3,104

- .26 -.22 .00

rating)

rating)

Sensitivity components entered stepwise, for sample size and number of independent *a* pc.001. p<.os. ** pc.01.

global rating variables.

with

forced

entry.

R’adi is R’

ANTECEDENTS

OF INFANT

REUNION

BEHAVIORS

113

havior was predicted by physical contact and responsiveness in combination. The global rating of sensitivity significantly added to the prediction Of resistant, but not avoidant, reunion behavior. None of the temperamental dimensions was significantly correlated with either avoidant or resistant behavior. The product-moment correlations ranged between - .14 and .lO, with about 80% of the coefficients being .05 or below. To investigate the possibility of interactive influences of infant temperament and maternal behavior, multiple regression analyses were performed with forced entry of first a maternal variable, second a temperamental variable, and third the interaction of the maternal and the temperamental variables (Cohen & Cohen, 1983). Thus, for each maternal variable, six analyses were made, one for each temperamental variable. The maternal variables used were the general rating and the different sensitivity components (Physical Contact, Responsiveness, and Intrusiveness). The interaction between the temperamental dimension Intensity/Activity and the maternal variable Physical Contact was found to yield a significant contribution. The already significant prediction of avoidant infant behavior from maternal physical contact behavior was significantly improved W change = .11, p C .OOl). The equation yielded a multiple correlation of .43, F(3,105) = 7.94, p < .OOl. The interaction effect was interpreted according to the procedures suggested by Cohen and Cohen (1983), which showed that the negative relationship between maternal physical contact and infant avoidance held primarily for infants high in intensity/activity. To further illustrate this result the correlations between Cmonth physical contact and la-month avoidance were calculated separately for infants above and below the median in the scale of Intensity/Activity in the 4-month questionnaire data. For the group high in intensity and activity, the obtained correlation was - 46 (p< JOI), whereas for the group low in intensity and activity it was - .09. In the analyses UShg SeIlSitiVity, intrusiveness or responsiveness as maternal variable, no interaction with temperament gave a significant contribution to the prediction of avoidant or resistant behavior. DISCUSSION In the present study, it proved fruitful to distinguish between various aspects of maternal sensitivity. The a priori defined components, responsiveness, physical contact, and intrusiveness, all contributed significantly in explaining the variance of the global rating of sensitivity and at the same time showed differential predictive relationships to infant avoidant and resistant behaviors observed 8 months later. It should also be noted, however, that the three components did not exhaustively describe the global aspect of maternal sensitivity; a large proportion of its variance was unexplained by the three components. The finding that the global rating contributed to the prediction of future infant resistant behavior, even when the separate components had been entered first in the regression analysis, also indicated that important aspects of the sensi-

114

BOHLIN.

HAGEKULL,

GERMER,

ANDERSON.

AND

LINDBERG

tivity construct were not tapped by the specific behaviors studied here. One aspect of maternal interaction which is theoretically strongly implicated as belonging to global sensitivity is responsiveness to infant crying and distress. In the present study such behaviors were observed, but not included, in the analyses because of the low frequency of infant distress in the observation situations. However, for the cases where such behaviors were present, the correlations between the global rating of sensitivity and the specific items “latency of response to distress” and “efficiency of comforting” were r(38) = .66 and r(41) = .67, respectively, both p< .OOl. Thus, responsiveness to distress signals might account for parts of the unexplained variance in the global sensitivity rating. None of the other observed maternal behaviors seemed to be closely related to the sensitivity construct. A factor analysis performed on all observed behaviors revealed two distinct varimax-rotated factors, one of which comprised the general sensitivity rating and the a priori defined sensitivity components. The other, uncorrelated, factor was composed of items measuring more quantitative aspects of maternal behavior and was interpreted as reflecting the mother’s involvement with her infant during the observations (unpublished data). Belsky et al. (1984) found no relation between amount of physical contact and attachment classifications as obtained in the standard Strange Situation and thus failed to replicate Ainsworth’s original observations concerning the antecedents of anxious/avoidant attachment (see Ainsworth et al., 1978). They attributed their negative results to the fact that their measure of physical contact did not give enough emphasis to affectionate behavior. Our positive finding was obtained with a measure of physical contact that did tap such qualitative aspects and is therefore in line with their reasoning. The variable Intrusiveness encompassed maternal behaviors often described as overstimulation. Belsky et al. (1984) presented data on the relation between reciprocal interaction and attachment patterns which were interpreted to mean that overstimulation is connected with the avoidant rather than the resistant pattern. In comparing the results from the two studies, methodological differences should be noted. More important, however, in understanding the present results for the Intrusiveness variable and relating them to the notion of overstimulation, is that some thought should be given to the situational generalizability of the behaviors in question. The behaviors subsumed in the concept of intrusiveness in the present study are perhaps not only shown by mothers who are generally active and overstimulating but also by mothers who know that they are being observed and want to be seen as “good mothers,” and, therefore, they overdo the interaction. If so, the behavior could be expected to be generally more pronounced in the face-to-face than in the diaper change situation, because demands on interactive behavior were clearly higher there than in the routine-care situation. It would also be assumed that the prediction from intrusiveness would primarly

ANTECEDENTS

OF INFANT

REUNION

BEHAVIORS

115

hold for observations in the face-to-face situation. Both these assumPtions were borne out by the data. Significantly more intrusiveness was observed in the face-to-face situation, t( 100) = 6.24, p < .OOl . Regression analyses with predictor variables taken separately from the face-to-face and diaper change situations in prediction of 1Zmonth resistant behavior showed that only in the former case was the regression equation for maternal intrusiveness in predicting resistant infant behavior significant, F(1,89) =9.57, p< .005, R = .31. Therefore, there is reason to believe that intrusive maternal behavior may be situation specific and should not be interpreted in terms of general overstimulation. It seems more likely that our observations of intrusiveness reflected on inconsistent kind of mothering, where the mother’s interactive efforts were more geared by her own needs to satisfy herself or others than by the infant’s. The pattern of findings for maternal intrusiveness and later infant resistant behavior is therefore interpreted as suggesting that one antecedent of anxious/ ambivalent attachment might be an infant’s experiences with a mother who is insensitive to infant signals because she is behaving more in accordance with her own picture of good mothering than in accordance with what the infant signals. This interpretation is in line with the view that mothers of insecureresistant infants possess an appropriate behavioral repertoire but are inconsistent in applying it (Ainsworth et al., 1978). A discussion in terms of situational specificity could also be applied to the relationship between items tapping specific components of sensitivity and its global ratings. The data indicated that, depending on observation situation, the components were differentially important for the explanation of general sensitivity variation. This suggests that the extent to which the global rating of sensitivity reflects a specific behavior belonging to this domain depends on the demands of the situation where the observations are made. Together with the finding of differential relationships between the various components of sensitivity and the two categories of anxious reunion behaviors, these results point to the necessity of making observations of sensitivity in a variety of situations if the predictive power of the concept for both avoidant and resistant infant behavior is to be demonstrated. This issue ties into the argument by Rushton, Brainerd, and Pressley (1983) concerning the principle of aggregation, where it is stated that the demonstration of substantial correlations between behavioral development variables depends on the use of multiple measurements. Because in the present study assessments were brief and limited to only two situations, the obtained predicted relationships are likely to provide only minimal estimates of the “true” relationship. Thus, the actual proportion of predicted variance should be regarded as of less importance than the fact that it was statistically significant. The idea of a direct influence of early temperamental characteristics on later attachment behaviors did not receive support in the present data. This supports the reasoning within the attachment tradition in which infant character-

116

BOHLIN,

HAGEKULL,

GERMER,

ANDERSON,

AND

LINDBERG

istics are subsumed into the caregiver-infant relationship in the sense that the sensitive caregiver adjusts to the individual characteristics of her infant (Sroufe, 1985). However, the interactionistic reasoning within temperament research (e.g., Goldsmith & Alansky, 1987) received some support in that our results suggested that the temperamental dimension of Intensity/Activity in interaction with a specific caregiver behavior component, physical contact, might play a role in explaining the development of avoidant behavior. Highly intensive and active infants receiving little emotional physical contact would seem to be at risk for developing avoidant behavior. This finding was not predicted and appeared in a large set of analyses and should therefore be interpreted with caution. In discussing the predictive findings of the present study, infant resistant and avoidant reunion behaviors have not been equated with the two patterns of anxious attachment. However, the results are congruent with the emphasis of traditional attachment theory on infants’ interactive experiences as antecedents of security of attachment. They also suggest that the two patterns of anxious attachment are differentially related to different aspects of maternal sensitivity. Above all, the results call for an increased interest, both in theoretical and empirical contexts, in the various behavioral aspects subsumed in the concept of maternal sensitivity. REFERENCES Ainsworth, M., Bell, S., & Stayton, D.J. (1971). Individual differences in Strange Situation behavior of one-year-olds. In H.R. Schaffer (Ed.), The origins of human social relations. London: Academic. Ainsworth, M., Blehar, M., Waters, E., & Wall, S. (1978). Pafterns of attachmenf. Hillsdale, NJ: Erlbaum. Belsky, J., & Rovine, M. (1987). Temperament and attachment security in the Strange Situation: An empirical rapprochement. Child Development, 58, 787-795. Belsky, J., Rovine, M., & Taylor, D.G. (1984)-The Pennsylvania Infant and Family Project, III: The origins of individual differences in infant-mother attachment: Maternal and infant contributions. Child Development, 55, 718-728. Blehar, M.C., Lieberman, A.F., & Ainsworth, M.D.S. (1977). Early face-to-face interaction and its relation to later infant-mother attachment. Child Development, 48, 182-194. Bohlin, G., & Hagekull, B. (1987). Good mothering: Maternal attitudes and mother-infant interaction. Infant Mental Health Journal, 8, 352-363. Bohlin, G., Hagekull, B., & Lindhagen, K. (1981). Dimensions of infant behavior. Infant Behavior and Development, 4, 83-96. Bowlby, J. (1969). Attachment and loss. Vol. I: Attachment. New York: Basic Books. Carey, W.B. (1970). A simplified method for measuring infant temperament. Pediatrics, 77, 188-194. Cohen, J., & Cohen, P. (1983). Applied multiple regression/correlation analysis,for the behavioral sciences. Hillsdale, NJ: Erlbaum. Goldsmith, H.H., & Alansky, J.A. (1987). Maternal and infant temperamental predictors of attachment: A meta-analytic review. Journal of Consulting and Clinical Psychology, 55, 805-816.

ANTECEDENTS

OF INFANT

REUNION

117

BEHAVIORS

Hagekuh, B. (in press). Longitudinal stability of temperament within a behavioral style framework. in G.A. Kohnstamm, J. Bates, & M.K. Rothbart (Ed.), Childhood remperomenl. A forum on key issues and problems. New York: Wiley. Hagekull, B., & Bohlin, G. (1981). Individual stability in dimensions of infant behavior. 1/anl Behavior and Development, 4, 97-108. Hagekull, B., Bohlin, G., & Lindhagen, K. (1984). Validity of parental reports. Infonf Behavior and Development, 7, 77-92. Lamb, M.E., Thompson, R.A., Gardner, W., & Charnov, E. L. (1985). Infont-mofher ottachmerit: The origins and developmental significance of individual differences in Strange Situofion behavior. Hillsdale, NJ: Erlbaum. Main, M., & Weston, D. R. (1982). Avoidance of the attachment figure in infancy: Description and interpretations. In C.M. Parkes & J. Stevenson-Hinde (Eds.), Thepluce of attachment in human behavior. New York: Basic Books. Rushton, J.P., Brainerd, C.J., & Pressley, M. (1983). Behavioral development and construct validity: The principle of aggregation. Psychological Bulletin, 94, 18-38. Sroufe, L.A. (1985). Attachment classification from the perspective of infant-caregiver relationships and infant temperament. Child Development, 56, l-14. Stevenson, M.B., & Lamb, M.E. (1979). Effects of infant sociability and the caretaking environment on infant cognitive performance. Child Development, SO, 340-349. Thomas, A., Chess, S., Birch, H.G., Hertzig, M., & Korn, S., (1963). Behavioral individuality in early childhood. New York: New York University Press. Vaughn, BE, Tamldson, B., Crichton, L., & Egeland, B. (1980). Relationships between neonatal behavioral organization and infant behavior during the first year of life. Infont Behavior and Development, 3, 47-66. 12 July

1988;

Revised

22 November

1988

n